Installation¶

from PyPI¶

pip install engram-memory-system

from source¶

git clone https://github.com/raya-ac/engram.git
cd engram
python3 -m venv .venv
source .venv/bin/activate
pip install -e .

requires python 3.11+. first run downloads two small models (~100MB total):

BAAI/bge-small-en-v1.5 (33MB) — embeddings
cross-encoder/ms-marco-MiniLM-L-6-v2 (22MB) — reranking

optional: API embedding backends¶

use cloud embedding APIs for higher quality:

pip install engram-memory-system[voyage]   # voyage-3.5, voyage-3.5-lite
pip install engram-memory-system[openai]   # text-embedding-3-small/large
pip install engram-memory-system[gemini]   # gemini-embedding-001
pip install engram-memory-system[api]      # all three

set API keys:

export VOYAGE_API_KEY="your-key"    # https://dash.voyageai.com/
export OPENAI_API_KEY="your-key"
export GEMINI_API_KEY="your-key"

see Embedding Backends for model comparison and switching.

docker¶

git clone https://github.com/raya-ac/engram.git
cd engram
docker compose up -d
# → http://localhost:8420

see Docker Guide for configuration.

build the ANN index¶

after installing, build the HNSW index for fast dense search:

engram index rebuild

this auto-updates on write/forget. only needed once on first install or after bulk operations.

verify¶

engram status

should show your database path, memory counts, and ANN index status.